Is the error "10055 no buffer space available" related to UDP ports?

Question

_{Tuesday, March 11, 2014 10:19 AM | 1 vote}

Hi all,

I am facing a "10055 no buffer space available" error on a host with Windows Server 2012. This error is actually reported on a SAP system on the host and it will cause the SAP system to be inaccessible.

We've reported this to SAP and after running the netstat command, they are claiming that the issue is down to an OS process/service which is taking up a lot of UDP ports. This has been identified as the WinRM service responsible for Remote Management.

Are the UDP ports related to the initial error? If yes then can anything be done to the WinRM service?

Thanks and regards.

All replies (7)

_{Wednesday, March 12, 2014 6:55 AM}

Hi,

An operation on a socket or pipe was not performed because the system lacked sufficient buffer space or because a queue was full. WinSock was unable to allocate additional memory to accommodate the function request. This error indicates a shortage of resources on your computer. It can occur if you are trying to run too many applications (of any kind) simultaneously on your computer. If this error tends to occur when you run certain applications, it might indicate that the applications do not properly return system resources (such as memory). It might also indicate that you are not closing the applications correctly.

Run netstat with parameter –o, combine with task manager, you can find out the service that is running out server resource. Close these applications. If this error persists, exit Windows or reboot your computer to remedy the problem.

Hope this helps.

_{Thursday, March 13, 2014 2:23 AM}

Hi,

The error can be resolved with a server restart but it keeps occurring on a weekly basis.

I can see from Netstat that the WinRM service is occupying a lot of UDP ports. Is this WinRM service really the cause of the error and if yes what can I do?

Thanks and regards.

_{Tuesday, March 18, 2014 2:35 AM}

Hi all,

We've had an issue on our server where a buffer space error (10055) keeps happening due to too many ports being utilized.

From Netstat, we can see that a lot of the UDP ports are being occupied by these services (NlaSvc, CryptSvc, Dnscache, LanmanWorkstation) which all share the same PID. Is it normal for this to happen? If not then what could be the cause?

Thanks and best regards.

_{Wednesday, March 19, 2014 3:33 AM}

Hi,

Sorry for the delay, since it is still the same question. We shall discuss here.

WinRM 1.1 and earlier:  The default HTTP port is 80.

WinRM 2.0:  The default HTTP port is 5985.

So I don’t think it is the root cause of this issue. For test, you can stop the service and check if resource could be released.

In addition, could you share a screenshot of netstat –o and task manager with the same PID.

At last, do you have any anti-virus software installed? If so disable it and try again.

Hope this helps.

_{Monday, March 24, 2014 4:18 AM}

Hi Daniel,

Can you provide me a way to share the netstat file with you as there are too many entries. I'm unable to post links here.

I've actually disabled the WinRM service but now there are other services (NlaSvc, CryptSvc, Dnscache, LanmanWorkstation) occupying the ports instead and they all share the same process. It should be quite abnormal for this to occur so I hope that you could help to identify a potential root cause.

There is also no anti virus software on the server.

Thanks and regards.

_{Friday, July 25, 2014 11:38 PM}

Any update on this?  I've got 3 servers doing the exact same thing - all UDP source ports are being consumed by:

svchost.exe                    928 CryptSvc, Dnscache, LanmanWorkstation,
                                   NlaSvc, WinRM

_{Friday, January 8, 2016 11:22 PM}

Sorry for dredging up an old issue, but I'm wondering if you ever found a resolution for this issue.  We've had the same problem on three of our Win 2012 R2 SAP servers over the past few weeks.  In our case there isn't anything obvious using up the ports--e.g. we have ~300 connected endpoints with ~100-200 active/time-wait sessions.

I am suspicious that it is an SAP process which is rapidly opening/closing UDP ports and exhausting temporarily exhausting the network resources, but all documentation I have found refers to this problem in relation to TCP sessions, not UDP.